Field experiment was performed from August 1996 to December 1999 to examine the feasibility of constructed wetland system for sewage treatment in rural areas. A pilot system was installed in Konkuk University and the effluent of septic tank for school building was used as an influent to the wetland treatment basin. The system was composed of sand and reed, and operated continuously including winter time. Average removal rate of about 70% was observed for BOD, COD, and SS, about 50% for T-P, and about 25% for T-N. The reason for poor T-N removal might be due to high loading rate and short retention time. The system demonstrated satisfactory effluent concentration and stable performance in growing season. And it also worked adequately in wintertime even below $10^{circ}C$ without freezing, and removal was still significant. The amount removed in BOD, COD, and SS was almost the same as in the growing season, and the amount removed in nutrients was about half of the one in growing season. Overall performance of the experimental system was compared with existing data base (NADB, 1994), and it was within the range of general system performance. As study period increased, removal rates for BOD, COD, SS, and T-P were consistently maintained and even enhanced, but removal rate for T-N decreased slightly. Wetland system was thought to be a feasible alternative for sewage treatment in rural area considering its low cost and low maintenance requirement. However, the effluent of the experimental wetland system often exceeded current effluent water quality standards, therefore, further treatment could be required if the effluent should be discharged to public waters. Wetland system of interest locates in rural area and is a part of rural ecosystem, therefore, ultimate disposal of reclaimed sewage for agricultural purpose or subsequent land treatment might be available and further research in this matter is recommended.